Hybrid pore-network and lattice-Boltzmann permeability modelling accelerated by machine learning

Rabbani, Arash; Babaei, Masoud

doi:10.1016/j.advwatres.2019.02.012

Cited by 125 publications

(48 citation statements)

References 69 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Hydraulic permeability in a throat can be simply calculated by Hagen-Poiseuille law (Schmid and Henningson 1994), and it is equal to r 2 ∕8 that r is the throat hydraulic radius. A more realistic approach to calculate the absolute permeability of a throat with an arbitrary cross-sectional shape is presented by Rabbani and Babaei (2019) and it is based on the Lattice-Boltzmann method (LBM) to simulate a steady-state single-phase flow through the throat. They showed that numerical results obtained for permeability of arbitraryshaped throats are highly correlated with averaged distance map of the throat image and permeability can be obtained as the following quadratic form:…”

Section: Extraction Of the Single Pore Network Modelmentioning

confidence: 99%

“…3b) and to obtain D , we perform an averaging on the non-zero values of a distance map. More details of statistical and physical justifications for this empirical equation are provided in Rabbani and Babaei (2019). Also, the basics and modelling assumptions of LBM simulation used in Rabbani and Babaei (2019) and extended to the current study are described in "Appendix 2".…”

Section: Extraction Of the Single Pore Network Modelmentioning

confidence: 99%

“…More details of statistical and physical justifications for this empirical equation are provided in Rabbani and Babaei (2019). Also, the basics and modelling assumptions of LBM simulation used in Rabbani and Babaei (2019) and extended to the current study are described in "Appendix 2". By describing the hydraulic behaviour of the links in a single continuum PNM, we are able to model fluid flow through the extracted networks that consist of meso-pores and meso-throats.…”

Section: Extraction Of the Single Pore Network Modelmentioning

confidence: 99%

“…We have defined the convergence criteria when calculated permeability value is stabilized and its relative error compared to the previous time step is less than 10 −6 . More details on LBM permeability calculation is provided in Rabbani and Babaei (2019). Freeman et al 2011).…”

Section: Appendix 2: Lattice-boltzmann Simulationmentioning

confidence: 99%

See 3 more Smart Citations

A Triple Pore Network Model (T-PNM) for Gas Flow Simulation in Fractured, Micro-porous and Meso-porous Media

2020

Self Cite

View full text Add to dashboard Cite

In this study, a novel triple pore network model (T-PNM) is introduced which is composed of a single pore network model (PNM) coupled to fractures and micro-porosities. We use two stages of the watershed segmentation algorithm to extract the required data from semireal micro-tomography images of porous material and build a structural network composed of three conductive elements: meso-pores, micro-pores, and fractures. Gas and liquid flow are simulated on the extracted networks and the calculated permeabilities are compared with dual pore network models (D-PNM) as well as the analytical solutions. It is found that the processes which are more sensitive to the surface features of material, should be simulated using a T-PNM that considers the effect of micro-porosities on overall process of flow in tight pores. We found that, for gas flow in tight pores where the close contact of gas with the surface of solid walls makes Knudsen diffusion and gas slippage significant, T-PNM provides more accurate solution compared to D-PNM. Within the tested range of operational conditions, we recorded between 10 and 50% relative error in gas permeabilities of carbonate porous rocks if micro-porosities are dismissed in the presence of fractures.

show abstract

Section: Extraction Of the Single Pore Network Modelmentioning

confidence: 99%

Section: Extraction Of the Single Pore Network Modelmentioning

confidence: 99%

Section: Extraction Of the Single Pore Network Modelmentioning

confidence: 99%

Section: Appendix 2: Lattice-boltzmann Simulationmentioning

confidence: 99%

See 2 more Smart Citations

A Triple Pore Network Model (T-PNM) for Gas Flow Simulation in Fractured, Micro-porous and Meso-porous Media

2020

Self Cite

View full text Add to dashboard Cite

show abstract

“…The input and output parameters are the second-order norms of SH frequencies of 1 to 15 and R inc at different RLS, respectively. The size of the hidden layer is set to be two-thirds of the sum of input and output parameters [52] to accelerate the training and avoid over-fitting. Table 3 depicts the training performance of ANN with mean squared error (MSE) and regression R value.…”

Section: Estimating R Inc Using Artificial Neural Network (Ann)mentioning

confidence: 99%

A Novel Multi-Scale Particle Morphology Descriptor with the Application of SPHERICAL Harmonics

2020

View full text Add to dashboard Cite

Particle morphology is of great significance to the grain- and macro-scale behaviors of granular soils. Most existing traditional morphology descriptors have three perennial limitations, i.e., dissensus of definition, inter-scale effect, and surface roughness heterogeneity, which limit the accurate representation of particle morphology. The inter-scale effect refers to the inaccurate representation of the morphological features at the target relative length scale (RLS, i.e., length scale with respective to particle size) caused by the inclusion of additional morphological details existing at other RLS. To effectively eliminate the inter-scale effect and reflect surface roughness heterogeneity, a novel spherical harmonic-based multi-scale morphology descriptor Rinc is proposed to depict the incremental morphology variation (IMV) at different RLS. The following conclusions were drawn: (1) the IMV at each RLS decreases with decreasing RLS while the corresponding particle surface is, in general, getting rougher; (2) artificial neural network (ANN)-based mean impact values (MIVs) of Rinc at different RLS are calculated and the results prove the effective elimination of inter-scale effects by using Rinc; (3) Rinc shows a positive correlation with the rate of increase of surface area RSA at all RLS; (4) Rinc can be utilized to quantify the irregularity and roughness; (5) the surface morphology of a given particle shows different morphology variation in different sections, as well as different variation trends at different RLS. With the capability of eliminating the existing limitations of traditional morphology descriptors, the novel multi-scale descriptor proposed in this paper is very suitable for acting as a morphological gene to represent the multi-scale feature of particle morphology.

show abstract

Photoannealing of Microtissues Creates High‐Density Capillary Network Containing Living Matter in a Volumetric‐Independent Manner

Schot,

Becker,

Paggi

et al. 2024

Advanced Materials

View full text Add to dashboard Cite

The vascular tree is crucial for the survival and function of large living tissues. Despite breakthroughs in 3D bioprinting to endow engineered tissues with large blood vessels, there is currently no approach to engineer high‐density capillary networks into living tissues in a scalable manner. We here present photo‐annealing of living microtissues (PALM) as a scalable strategy to engineer capillary‐rich tissues. Specifically, in‐air microfluidics was used to produce living microtissues composed of cell‐laden microgels in ultra‐high throughput, which could be photo‐annealed into a monolithic living matter. Annealed microtissues inherently give rise to an open and interconnected pore network within the resulting living matter. Interestingly, utilizing soft microgels enables microgel deformation, which leads to the uniform formation of capillary‐sized pores. Importantly, the ultra‐high throughput nature underlying the microtissue formation uniquely facilitates scalable production of living tissues of clinically relevant sizes (>1 cm3) with an integrated high‐density capillary network. In short, PALM generates monolithic, microporous, modular tissues that meet the previously unsolved need for large engineered tissues containing high‐density vascular networks, which is anticipated to advance the fields of engineered organs, regenerative medicine, and drug screening.This article is protected by copyright. All rights reserved

show abstract

Hybrid pore-network and lattice-Boltzmann permeability modelling accelerated by machine learning

Abstract: Hybrid pore-network and Lattice-Boltzmann permeability modelling accelerated by machine learning.

Cited by 125 publications

References 69 publications

A Triple Pore Network Model (T-PNM) for Gas Flow Simulation in Fractured, Micro-porous and Meso-porous Media

A Triple Pore Network Model (T-PNM) for Gas Flow Simulation in Fractured, Micro-porous and Meso-porous Media

A Novel Multi-Scale Particle Morphology Descriptor with the Application of SPHERICAL Harmonics

Photoannealing of Microtissues Creates High‐Density Capillary Network Containing Living Matter in a Volumetric‐Independent Manner

Contact Info

Product

Resources

About